Bidirectional modulation of synaptic transmission by insulin-like growth factor-I

Insulin-like growth factor-I (IGF-I) plays a key role in the modulation of synaptic plasticity and is an essential factor in learning and memory processes. However, during aging, IGF-I levels are decreased, and the effect of this decrease in the induction of synaptic plasticity remains unknown. Here we show that the induction of N-methyl-D-aspartate receptor (NMDAR)-dependent long-term potentiation (LTP) at layer 2/3 pyramidal neurons (PNs) of the mouse barrel cortex is favored or prevented by IGF-I (10 nM) or IGF-I (7 nM), respectively, when IGF-I is applied 1 h before the induction of Hebbian LTP. Analyzing the cellular basis of this bidirectional control of synaptic plasticity, we observed that while 10 nM IGF-I generates LTP (LTPIGF-I) of the post-synaptic potentials (PSPs) by inducing long-term depression (LTD) of the inhibitory post-synaptic currents (IPSCs), 7 nM IGF-I generates LTD of the PSPs (LTDIGF-I) by inducing LTD of the excitatory post-synaptic currents (EPSCs). This bidirectional effect of IGF-I is supported by the observation of IGF-IR immunoreactivity at both excitatory and inhibitory synapses. Therefore, IGF-I controls the induction of Hebbian NMDAR-dependent plasticity depending on its concentration, revealing novel cellular mechanisms of IGF-I on synaptic plasticity and in the learning and memory machinery of the brain.


Introduction
Insulin-like growth factor-1 (IGF-I) is a peptide with well-known trophic functions in the brain, but its neuromodulatory role is still under study.It is well documented that IGF-I regulates neuronal firing (Carro et al., 2000;Nuñez et al., 2003;Gazit et al., 2016), and modulates synaptic transmission in several areas of the central nervous system (CNS), such as hippocampus or cerebellum (Nilsson et al., 1988;Araujo et al., 1989;Castro-Alamancos and Torres-Aleman, 1993;Seto et al., 2002;Fernando Maya-Vetencourt et al., 2012).IGF-I is also able to enhance glutamatergic synaptic transmission in rat hippocampal slices of juvenile

Materials and methods
All animal procedures were approved by the Ethical Committee of the Universidad Autónoma of Madrid, and Cajal Institute and are in accordance with Spanish (R.D. 53/2013) and European Community Directives (2010/63/EU), which promote the animal welfare.Male C57BL/6J mice were housed in standard laboratory cages with ad libitum access to water and food in temperature-and humiditycontrolled rooms under a 12 h/12 h light/dark cycle with up to five animals per cage and were used for slice electrophysiology.Mice from different litters were used to increase the reproducibility of the experiments.

Spike-timing dependent plasticity, Hebbian LTP
In some experiments, after applying IGF-I as described above, spike-timing dependent plasticity (STDP) was induced (Figure 1).Induction of STDP was achieved by pairing pre-and postsynaptic action potentials 10 ms away.Presynaptic action potentials were evoked by electrical stimulation of basal afferent inputs.Stimulus intensity was adjusted to obtain 3-5 mV PSPs responses.Postsynaptic action potentials were elicited by a brief current injection through the recording pipette (5 ms, 200-400 pA).Basal PSPs were recorded for 10 min before the pre-post associations were induced (30, 20, 50 and 100 pairings were studied).After that, PSPs were recorded for another 50-60 min.Amplitude of PSPs 5 min before (−5 to 0 min interval) and 50 min after (45 to 50 min interval) the LTP H induction protocol were measured to compare the extent of the PSP potentiation.

IGF-I effects on synaptic transmission recordings
In all experiments, basal postsynaptic currents (PSCs) were recorded for 5 min at 0.33 Hz (voltage-clamp configuration), then basal postsynaptic potentials (PSPs) were recorded at 0.2 Hz

Synaptic stimulation increased
Synaptic stimulation increased (SSI), after 5 min of stable baseline of EPSCs in voltage-clamp, the recording was switched to the current-clamp and the stimulation intensity was increased to values in which an EPSP followed by an AP was triggered during 15 min at 0.2 Hz.Next, the values of synaptic stimulation were returned to control conditions and PSCs were recorded for 25 min (Noriega-Prieto et al., 2019) (Figure 4).

Simulated spike
Simulated spike, after 5 min of stable baseline of EPSCs in voltageclamp, after that, recording mode was switched to the current-clamp and a previously obtained neuronal spike was used as stimulus to depolarize the neurons through the recording pipette during 15 min at 0.2 Hz.Next, we were returned to control conditions and EPSCs were recorded for 25 min.
For IGF-IR immunogold labeling, sections containing the somatosensory cortex were used.Sections were first washed with PBS and incubated in a 50 mM glycine solution 5 min to increase antibody binding efficiency.Following a standard immunocytochemical protocol, tissue was first free-floating incubated in a rabbit polyclonal anti-IGF-IRα antibody (1/250, Santa Cruz) in a PBS 0.1 M/1% BSA solution for 48 h at 22°C.Then, sections were washed in PBS, and incubated with 1.4 nm gold-conjugated goat anti-rabbit IgG (1:100; Nanoprobes) overnight at 22°C.After post-fixing with 2% glutaraldehyde and washing with 50 mM sodium citrate, labeling was enhanced with the HQ Silver™ Kit (Nanoprobes), and gold toned.Finally, immunolabeled sections were fixed in 1% osmium tetroxide, block stained with uranyl acetate, dehydrated in acetone, and flat embedded in Araldite 502 (EMS, United States).Selected areas were cut in ultrathin sections (70-80 nm) and examined and photographed with a JEOL JEM1400 electron microscope.As a control for the immunogold technique, sections were processed as above but omitting the primary antibody.No specific labeling was observed in these control sections.We identified excitatory and inhibitory synapses based on the morphological appearance of the PSD in EM images.Excitatory (asymmetric) synapses typically display a pronounced post-synaptic density (PSD) much thicker than the relatively faint presynaptic thickening, while the PSD in inhibitory (symmetric) synapses looks similar to the presynaptic membrane (any synapse with a less marked PSD, similar to the presynaptic thickening, was classified as "symmetric") (Gray, 1969).Immunohistochemistry C57BL6/J mice (15 days old) were anaesthetized with pentobarbital (60 mg/kg) and perfused transcardially with 4% paraformaldehyde in 0.1 M phosphate buffer pH 7.4 (PB).The brain was then removed and post-fixed overnight.Coronal 50 μm-thick sections were cut and collected in PBS.Free-floating brain sections were pre-incubated in PB with Triton X-100 1% and bovine serum albumin 1%, and incubated overnight at 4°C with the respective primary antibody [mouse anti-PSD95 (1:500), mouse anti-vGAT (1:50) and rabbit anti-IGF-IR (1:100)].Incubations were performed in two steps (first, PSD95/vGAT and then IGF1R).The following primary antibodies were utilized: a mouse anti-PSD95 antibody (Thermo Scientific 1:500), a mouse vGAT (Synaptic System 1:50) and a rabbit anti-IGF1R antibody (Santa Cruz, 1:100).Secondary Alexa-labeled antibodies from Molecular Probes (Donkey antimouse 488 Alexa-conjugated for the PSD95 and vGAT detection and donkey anti-mouse 594 Alexa-conjugated for the IGF1R detection) were used at a final concentration of 1:1000.All sections were counterstained with Hoechst (Calbiochem, 1:500).The incubation periods used were 1 h at room temperature (RT) and 48 h at 4°C for primary antibodies, 1 h RT and 24 h at 4°C for secondary antibodies, and 15 min for Hoechst incubation.

Data analysis
Pre-or post-synaptic origin of the synaptic plasticity induced by IGF-I was tested by analyzing the modification in the variance that parallel the synaptic current amplitude change, which reflect the change in transmitter release probability (Clements, 1990;Malinow et al., 1990;Kuhnt and Voronin, 1994).To estimate the modification of the synaptic current variance, we first calculated the noise-free coefficient of variation (CV NF ) of the synaptic responses before (control conditions) and during IGF-I.We used the formula CV NF = √ (δ synaptic current 2 − δ noise 2 )/m, where δ noise 2 and δ synaptic current 2 are the baseline and synaptic current peak variance, respectively, and m is the mean peak amplitude of the synaptic current.The ratio of the CV NF (CV R ) before (control conditions) and during IGF-I was obtained then for each neuron as CV IGF-I /CV control (Clements, 1990).We constructed plots comparing variation in the normalized m (termed M) to the change in 1/CV R 2 in each cell (Malinow et al., 1990).In these plots, depression of the synaptic currents has a presynaptic origin when values are below or in the diagonal, whereas points above the diagonal indicate a postsynaptic origin.However, potentiation of synaptic currents has a presynaptic origin when values are above or in the diagonal, whereas points below the diagonal indicate a postsynaptic origin (Faber and Kornt, 1991).This method requires a binomial EPSC amplitude distribution, a condition that must be met for the synaptic variance to reflect the probability of transmitter release.We could not directly test whether our data fitted a binomial distribution, but synaptic fluctuations were always evident and we assumed that synaptic release followed a binomial distribution.Data analysis was done in Clampfit 10 (Axon Instrument) and graphs were drawn in SigmaPlot 11.In all cases, statistical estimates were made with student's two-tailed paired t-tests, and data are presented as means ± SE.To analyze the effects of different treatments and conditions, we carried out multiple comparison testing between the different groups using one-way ANOVA.The post hoc test used was Holm-Sidak or Dunn's for parametric or non-parametric values respectively, versus control comparisons using the "after stimulus" situation and the "basal" condition as the controls to compare with a Holm-Sidak's or Dunn's multiple comparisons test.Statistical differences were established with p < 0.05 ( * ), p < 0.01 ( * * ) and p < 0.001 ( * * * ) for student's t-test and p < 0.05 (#), p < 0.01 (##) and p < 0.001 (###) for the post hoc Holm-Sidak and Dunn's test.The sample size is shown as n = x/y (x shows the number of cells, and y shows the number of animals).

IGF-I regulates the induction of Hebbian LTP (LTP H ) in a concentration dependent manner
We have previously shown that 10 nM IGF-I decreases the induction threshold of LTP H (Noriega-Prieto et al., 2021).Here, we analyzed the effect of a slightly lower dose of IGF-I on LTP H induced by spike timing dependent plasticity (STDP) at layers 2/3 of the barrel cortex (Figure 3A).We first checked in control conditions, the STDP protocol, consisting in a subthreshold PSP followed by a back-propagating action potential (BAP) at delays of 10 ms repeated 50 times at 0.2 Hz (Figure 3B) and then we performed similar experiments in slices in which previously IGF-I at doses of 10 nM, 7 nM or 5 nM was bath applied during synaptic stimulation of layer 4 at 0.2 Hz (Figure 3C).In control conditions, the STDP protocol induced an LTP H of PSPs (from 96.01 ± 4.12 to 137.62 ± 8.33% of amplitude, p < 0.001; n = 7/3; Figures 3D,G, LTP control).However, in slices pretreated with 10 nM IGF-I, the STDP protocol induced a more robust LTP H of PSPs (from 99.36 ± 1.08 to 141.59 ± 4.38% of amplitude after 20 pairings, p < 0.001; n = 6/2 and from 99.36 ± 0.89 to 169.13 ± 6.36% of amplitude after 50 pairings, p < 0.001; n = 6/3; Figures 3E,G, LTP facilitated).Conversely, in slices pretreated with 7 nM IGF-I, the STDP protocol did not induce LTP H of the PSPs (from 101.66 ± 1.12 to 99.85 ± 6.35% of amplitude after 50 pairings, p > 0.05; n = 6/3; Figures 3F,G, LTP impairment).Interestingly, 7 nM IGF-I with 100 pairings repetition induced an LTD (from 102.74 ± 1.26 to 72.65 ± 5.19% of amplitude after 100 pairings, p < 0.001; n = 5/2; Figures 3F,G, LTP impairment), which further supports the idea of different IGF-I levels differently regulate the synaptic plasticity.Additionally, 50 pairings of pre-postsynaptic neuronal activity in the presence of picrotoxin (50 μM) and CGP (1 μM) to isolate the excitatory synaptic transmission, after 10 nM IGF-I did not induce any modulatory effect in the excitatory postsynaptic currents (EPSCs) (from 100.57 ± 0.63 to 105.25 ± 10.13% of amplitude after 50 pairings; p > 0.05, n = 6/3; Supplementary Figures S2A-D, 50 pairings).Thus, only 10 nM, but not 7 nM IGF-I favored the induction of LTP H through the modulation of inhibitory synaptic transmission (Noriega-Prieto et al., 2021).In other words, the activation of IGF-IRs favors or impairs LTP H depending on IGF-I concentration.

Bidirectional modulation of synaptic transmission by IGF-I levels
We next compared the effects of these two doses of IGF-I on synaptic transmission.For this set of experiments, we recorded at L2/3 We next analyzed the locus of expression of IGF-I mediated LTD of the EPSCs.We first studied whether it was paralleled by a decrease in the probability of glutamate release.We pharmacologically isolated the EPSCs by blocking GABA A inhibition with PiTX, and applied 7 nM IGF-I (Supplementary Figure S1).Under these conditions, IGF-I induced a similar LTD of the ESPCs than IGF-I 7 nM, that lasted at least 40 min of recording (from 99.59 ± 0.44 to 67.36 ± 3.87% of amplitude, p < 0.001, n = 6/2; Supplementary Figure S1A, black circles).This effect was inhibited with the IGF-IR antagonist NPV-AEW 554 (from 100.37 ± 0.37 to 94.87 ± 3.67% of amplitude, p > 0.05, n = 5/2; Supplementary Figure S1A, white circles).To investigate the pre or postsynaptic locus of expression of this LTD, we analyzed the effect of IGF-I on the pair pulse ratio (PPR) and on the EPCS variance by measuring the coefficient of variation (CV).The effect of IGF-I on both the PPR (from 0.98 ± 0.01 to 1.28 ± 0.10 before and after IGF-I respectively, p < 0.05, n = 6/2; Supplementary Figure S1B left) and the analysis of 1/CV 2 plots (linear correlation R 2 = 0.974, Supplementary Figure S1B right) revealed that LTD IGF-I was due to a decrease in the probability of release of glutamate.

Cytosolic calcium levels determine the sign of IGF-I mediated synaptic plasticity
Neurons secrete IGF-I by an activity-dependent pathway of exocytosis, and a mild depolarization is sufficient to induce IGF-I secretion in olfactory bulb neurons (Cao et al., 2011).Therefore, we next tested whether changes in cytosolic Ca 2+ levels of postsynaptic PNs were involved in IGF-I potentiation of the EPSC.We carried out similar experiments as before, but in the presence of the Ca 2+ chelator BAPTA (20 mM) in the recording pipette (Figure 2A).Under these conditions, 10 nM IGF-I induced an EPSCs depression (43.90 ± 4.04% of baseline, p < 0.001, n = 6/3, Figure 2B, 10 nM IGF-I, green circles) rather than an EPSCs potentiation observed in the absence of BAPTA.This EPSC depression was absent with 7 nM IGF-I (from 98.79 ± 1.37 to 99.19 ± 2.23% of amplitude, p > 0.05, n = 6/3; Figure 2B, 7 nM IGF-I, blue circles).Moreover, the plasticity was abolished in the presence of the IGF-I receptor antagonist NPV-AEW 554 (from 99.42 ± 1.79 to 104.08 ± 4.39% of amplitude, p > 0.05, n = 5/2; Figure 2B, 10 nM IGF-I + NVP, light red circles).These results indicate that IGF-I mediate EPSCs potentiation or depression depending on the cytosolic Ca 2+ level, being all these forms of synaptic plasticity dependent on the activation of IGF-IRs.
Synaptotagmin 10 (Syt10) acts as the Ca 2+ -sensor that triggers IGF-I exocytosis in olfactory bulb neurons (Cao et al., 2011).Thus, we prevented exocytosis by using the light chain of the B type botulinum toxin (i.e., Botox 0.5 μM), which inhibits the SNARE protein-mediated membrane fusion of endosome complexes, and tested whether IGF-I effects depend on exocytosis.Surprisingly, 7 nM IGF-I did not modulate the EPSCs (data not shown), while IGF-I 10 nM depressed the EPSCs under BOTOX (42.26 ± 8.63% of baseline, p < 0.01, n = 6/3; Figure 2C, BOTOX + IGF-I 10 nM), suggesting that higher IGF-I concentrations are required for LTD IGF1 under Botox.Interestingly, increasing IGF-I to 20 nM was able to induce the potentiation of the EPSCs (227.80 ± 20.43% of baseline, p < 0.001, n = 7/3; Figure 2C, BOTOX + IGF-I 20 nM) indicating that higher IGF-I concentrations are required to LTP IGF1 under Botox probably because it blocked the activity-dependent release of IGF-I from the postsynaptic neuron.

Synaptic stimulation and spiking activity can induce IGF-I mediated synaptic plasticity
Since cytosolic calcium levels and exocytosis are determinant in the induction of LTD and LTP by IGF-I, we next tested whether increases of cytosolic calcium induced by synaptic stimulation could be enough to induce IGF-I-mediated synaptic plasticity.After 5 min of recording the EPSCs, we increased the intensity of synaptic stimulation (SSI, synaptic stimulation increase) until a PSP followed by an action potential was recorded (Figures 4A,B).Next, we maintained evoked these suprathreshold responses for 15 min by SSI, and then we turned the stimulation intensity back to control values.This protocol of stimulation induced a LTD of the EPSCs (from 101.35 ± 2.43 to 54.54 ± 8.43% of amplitude, p < 0.01, n = 5/2; Figures 4A,B, black circles, SSI) that was prevented with NVP (from 101.09 ± 0.91 to 94.33 ± 4.50% of amplitude, p > 0.05, n = 6/3; Figures 4A,B, white circles, SSI + NVP), or under Botox (from 100.33 ± 1.01 to 100.60 ± 2.43% of amplitude, p > 0.05, n = 6/3; Figures 4A,B, red circles, SSI + Botox).These results suggest that suprathreshold responses induced by SSI are sufficient to induce an IGF-I-mediated LTD of the EPSCs dependent on endosome exocytosis.Finally, we checked whether simulation of these suprathreshold responses (Figures 4C,D, Simulated Spike) by injecting them through the patch recording electrode could induce a similar LTD of the EPSCs.As shown in Figures 4C,D, simulation of suprathreshold responses for 15 min was enough to induce a presynaptic LTD of the EPSCs (40.56 ± 3.58% of baseline, p < 0.01, n = 5/2; Figures 4C-E purple circles, simulat.spike), that is prevented with NVP (3.25 ± 3.38% of baseline, p > 0.05, n = 5/2; Figures 4C,D blue circles, simulat.spike + NVP).These results suggest that

IGF-IR is present at presynaptic terminals of both excitatory and inhibitory synapses
We finally analyzed whether IGF-IRs were present at presynaptic terminals of excitatory in the postsynaptic L2/3 PNs.Electron micrographs demonstrated IGF-IR silver-enhanced immunogold labeling (Figures 5A,B, white arrows) at presynaptic terminals of both excitatory (Figure 5A) and inhibitory (Figure 5B

Role of IGF-I in the modulation of Hebbian synaptic plasticity
In the central nervous system, IGF-I/IGF-IR signaling is critical for experience-dependent synaptic and neuronal plasticity in sensory cortices, adult neurogenesis (LLorens-Martín et al., 2010), synaptic vesicle release, and neuronal excitability (Nuñez et al., 2003;Maglio et al., 2021).Our present results challenge the standard view that IGF-I favors cognition by inducing LTP in cortical circuits, and it expands the range of brain IGF-I actions.Thus, we provide novel evidence that IGF-I levels determine the sign of plasticity induced at the barrel cortex; high levels induce LTP, while lower levels induce LTD, which in turn, favor or impair Hebbian LTP, respectively (Figure 6).While the reduction in the inhibitory tone by 10 nM IGF-I and the subsequent LTP of the synaptic plasticity reduces the threshold for inducing Hebbian LTP,  the reduction of the excitatory transmission induced by 7 nM IGF1 would increase the threshold maybe by an insufficient activation of the necessary number of NMDARs to increase the cytosolic calcium to induce synaptic plasticity.Although the observation that IGF-I action on the brain is crucial for learning and memory is well established, here we are proposing an additional interpretation of the classical IGF-I mechanism.These results indicate that brain levels of IGF-I play an important role in synaptic plasticity, which is crucial for memory processes.Therefore, any changes in brain IGF-I levels could impact synaptic plasticity and cognitive functions.

Depression of the excitatory synapses by IGF1
Extracellular IGF-I activates IGF-IRs under resting conditions, maintaining basal transmission and ongoing spiking activity in the physiological range, and induces synaptic modulation (Gazit et al., 2016).Presynaptic IGF-IRs are basally active, thus regulating glutamatergic synaptic transmission by modulating the glutamate release probability (Gazit et al., 2016).In fact, it can be concluded that tonic release of IGF-I and subsequent activation of IGF-IRs modulates synaptic vesicle release, leading to a short term depression in excitatory hippocampal neurons (Gazit et al., 2016).In agreement with these results, we also observed a long-term depression of synaptic transmission in the absence of postsynaptic activity (when firing at the recorded PN was prevented in voltage clamp) or when BAPTA abolished the cytosolic calcium increase mediated by this activity at PNs.Under Botox 7 nM IGF-I is not sufficient to induce LTD and 10 nM IGF-I is required for the LTD induction, whereas increasing the concentration to 20 nM IGF-I induces LTP.This demonstrates that endosome fusion is critical for establishing the concentration threshold for the induction of LTD and LTP by IGF-I, and suggests that exogenous application of IGF-I could induce neuronal-released IGF-I stored in endosomes (a calcium and endosome dependent IGF-I induce IGF-I release).These results strongly indicate that the tonic modulation of synaptic transmission induced by IGF-I described at the hippocampus is also present at glutamatergic synapses of the barrel cortex.

IGF-I modulates both excitatory and inhibitory synaptic activity throughout the CNS
Our previous results show that 10 nM IGF-I induces an LTP of the PSPs, that we have termed LTP IGFI (Noriega-Prieto et al., 2021).Moreover, we demonstrated that IGF-IR activation favors the induction of NMDAR-dependent LTP and improves texture discrimination of the mouse whiskers.Here we show that lower IGF-I levels (7 nM) induce LTD of the PSPs, that we have termed LTD IGFI and impairs the induction of NMDAR-dependent LTP.These results are supported by the presence of IGF-IR in the presynaptic terminal of both excitatory and inhibitory synapses described in this work.In addition, based on the evidence, this mechanism of action has been studied in the hippocampus, where IGF-I can induce the release of GABA to regulate endogenous ACh release, possibly acting via GABAergic neurons (Seto et al., 2002).Moreover, olfactory learning in a social context selectively induces LTP of the GABAergic component of reciprocal synapses between granule and mitral cells in the medial olfactory bulb (MOB), requiring an autocrine and/or paracrine action of IGF-I to enhance postsynaptic GABA receptor function.Indeed, blocking Ca 2+ -triggered IGF-I release prevents GABAergic LTP (Liu et al., 2017).At any rate, our observations contribute to the notion that IGF-I modulates both excitatory and inhibitory synaptic activity throughout the CNS (Maglio et al., 2021;Noriega-Prieto et al., 2021).According to our results, IGF-I is able to induce a dual effect on glutamatergic synaptic transmission by the activation of IGF-IRs.In fact, both a pre-synaptically IGF-I-mediated potentiation and depression of the EPSCs were observed, and both were prevented by NPV-AEW 554 (Maglio et al., 2021;Noriega-Prieto et al., 2021).

Bidirectional modulation of synaptic transmission by IGF-I
On the other hand, we also observed that IGF-I induced an increase in glutamatergic synaptic transmission through the activation of IGF-IRs.As discussed above, IGF-I-induced potentiation of excitatory synaptic transmission is dependent on the presence of exogenously applied IGF-I, thereby pointing to the importance of reaching specific local levels of IGF-I.Indeed, only IGF-I mediated depression of the EPSCs are observed by the release of IGF-I induced by high synaptic stimulation or simulated spikes (Figure 4).These results points to the importance of IGF-I uptake from the plasma in the induction of bidirectional plasticity, since only depression of the synaptic transmission can be induced without extracellular perfusion of IGF-I.Whereas there is a tonic IGF-I release that induces depression of glutamatergic synaptic transmission (Gazit et al., 2016), our results suggest that the bidirectional effect of IGF-I on the modulation of the EPSCs may depend on the levels of IGF-I reached (Figure 6).An IGF-I -mediated EPSC depression is produced when PNs fires during suprathreshold responses, whereas an IGF-I-mediated EPSC potentiation is induced by bath applied IGF-I.Under physiological conditions, higher levels of IGF-I could be reached in a neuronal-activity dependent manner (Nishijima et al., 2010).In fact, we described that serum IGF-I input to the brain is regulated by an activity-driven process that includes increased blood-brain barrier permeability to serum IGF-I (Nishijima et al., 2010).Also, it has been demonstrated that physical exercise induces increased brain uptake of serum IGF-I by specific groups of neurons throughout the brain (Carro et al., 2000;Nuñez et al., 2003).This increase in the uptake of IGF-I, under both mentioned physiological circumstances, induces an increase in neuronal excitability, which perfectly correlates with our previous in vivo results (Noriega-Prieto et al., 2021).The ex vivo investigation showed here suggests that this increase in IGF-I uptake would be essential in EPSC potentiation by increasing local IGF-I levels to those required for the modulation of synaptic transmission, playing a role in the increase in cortical activity.
In summary, IGF-I induces bimodal regulation of the excitatory and inhibitory synaptic transmission depending on its levels.This bidirectional action probably contributes to favor or impair the generation of associative memories impacting the barrel cortex-related behaviors.Although there is a SNARE dependent release of IGF-I from neurons (Cao et al., 2011) and Botox inhibits the SNARE proteinmediated membrane fusion of endosome complexes, the application of Botox, though effective in blocking the release of IGF-I stored in endosomes, can affect other neuromodulators stored in endosomes that modulate synaptic transmission, such as BDNF (Nasrallah et al., 2021), and interfere with glutamate receptor trafficking (Pellett et al., 2015).Additionally, further studies are necessary to investigate whether the bidirectional modulation induced by IGF-I is maintained in older animals.Despite potential limitations, our previous findings, together with the present work, reveal novel insights into the mechanisms of IGF-I signaling in the cortex.

Significance statement
Insulin-like growth factor-I (IGF-I) signaling plays key regulatory roles in multiple processes of brain physiology, such as learning and memory, and brain pathology, such as Alzheimer disease.Yet, the underlying mechanisms remain largely undefined.Here we demonstrate that IGF-I signaling triggers long-term potentiation (LTP) or long-term depression (LTD) of synaptic transmission at cortical synapses in a concentration dependent manner, thus regulating the induction of Hebbian synaptic plasticity.The present work represents an important conceptual advance in our knowledge of the cellular basis of IGF-I signaling in brain function.

FIGURE 1
FIGURE 1 Hebbian LTP is favored and impaired by IGF-I 10 nM and 7 nM, respectively.(A) DIC image showing the recording (Rec.)located in layer II/III and the stimulation (Stim.)electrodes locate in layer IV in a slice (scale bar 100 μm).(B) Representative responses recorded (PSP followed by an AP with a 10 ms delay and a frequency of 0.2 Hz) during the STDP protocol in control.(C) Time course scheme showing the IGF-I exposure, washout and STDP induction (black arrow).(D) Superimposed representative PSPs (pre-pairing, black trace) and 40 min after 50 pairings (post 50 pairings, blue trace) in the control experiment of STDP.(E) Same as D but when 10 nM IGF-I was bath applied before the STPD.(F) Same as D but when 7 nM IGF-I was bath applied before the STPD.(G) Bar plot showing the PSP peak amplitude as percentage of the control before (pre-paring, blue bars) and 40 min after (post-pairing, red bars) the application of the STDP protocol in which 10, 20 and 50 parings were applied in control (LTP H -Control), or after 10 nM IGF-I (LTP H -facilitation) or after 10, 20, 50 and 100 parings 7 nM IGF-I (LTP H -impairment). *p < 0.05, **p < 0.01, and ***p < 0.001; student's paired t-test.# p < 0.01, ## p < 0.01, and ### p < 0.001; one-way ANOVA with post hoc Holm-Sidak; n.s., nonsignificant (p > 0.05).

FIGURE 2
FIGURE 2 LTP and LTD of the postsynaptic potentials are induced by IGF-I 10 nM and 7 nM, respectively.(A) Schematic representation of layer 2/3 pyramidal neuron recording (Rec.) and the stimulation electrode (Stim.) in the barrel cortex (scheme was performed with BioRender).(B top) Superimposed representative PSPs recorded before (black trace, control), during (red trace, IGF-I) and 40 min after 10 nM IGF-I washout (grey trace, washout), showing the induction of PSP potentiation by IGF-I (LTP IGF1 ).(B middle) Superimposed representative PSPs recorded before (black trace, control), during (red trace, IGF-I) and 40 min after 7 nM IGF-I washout (grey trace, washout), showing the induction of PSP potentiation by IGF-I (LTP IGF1 ).(B bottom) Superimposed representative PSPs recorded before (black trace, control), during (red trace, IGF-I) and 40 min after 5 nM IGF-I washout (grey trace, washout), showing the induction of PSP depression by IGF-I (LTD IGF1 ).(C) Bar plot showing the PSP peak amplitude as percentage of the control before (control) and 40 min after IGF-I (washout) 10 nM, 7 nM and 5 nM IGF-I.(D top) Superimposed representative PSCs recorded before (black trace, control), during (red trace, during IGF-I) and 40 min after IGF-I washout (grey trace, after IGF-I), showing the induction of PSP potentiation by 10 nM IGF-I, the PSP depression by IGF-I 7 nM and the no plasticity by IGF-I 5 nM.(D bottom) Time course of the PSC peak amplitude expressed as percentage of control before, during and after washing out the IGF-I 10 nM (white circles, short term potentiation STP) 7 nM (black circles, LTD) and 5 nM (grey circles, no plasticity, NP).(E) Bargraph showing the PSC change percentage showed in D bottom.*p < 0.05, **p < 0.01, and ***p < 0.001; student's paired t-test.# p < 0.01, ## p < 0.01, and ### p < 0.001; one-way ANOVA with post hoc Holm-Sidak; n.s., nonsignificant (p > 0.05).
FIGURE 3 IGF-I 10 nM induces LTD when postsynaptic calcium increases and endosome exocytosis are prevented by BAPTA and BOTOX, respectively.(A) Scheme showing the recording of layer 2/3 pyramidal neuron with a patch pipette containing 20 mM BAPTA and the stimulating electrode located at layer 4 of the barrel cortex (scheme was performed with BioRender).(B left) Time course of the EPSCs peak amplitude as percentage of the control under BAPTA in the patch pipette before, during (IGF-I, red empty bar) and after washout of 10 nM IGF-I (green circles), 7 nM IGF-I (blue circles) and in the presence of 10 nM IGF-I + NVP (orange circles).(B right) Bar plot showing the EPSC peak amplitude as percentage of the control under BAPTA in the patch pipette before (basal, grey bars) and after (light red bars) washout of 10 nM IGF-I, 7 nM IGF-I and 10 nM IGF-I + NVP.Note that under BAPTA LTD is induced by 10 nM IGF-I.(C left top) Superimposed representative EPSCs recorded before (control, black trace), during (IGF-I, red trace) and 40 min after washout of 10 nM IGF-I (washout, grey trace) with a BOTOX 1 μM containing intracellular solution in the patch pipette (Botox + IGF-I 10 nM).(C left bottom) Same as top but increasing IGF-I concentration to 20 mM.(C right) Bar plot showing the EPSC peak amplitude as percentage of the control under Botox in the patch pipette before (control, dark blue bars), during (IGF-I, middle blue bars) and 40 min after washout of IGF-I (washout, green bars) in normal ACSF (BAPTA bars) when applying IGF-I 10 nM (left bars, Botox + 10 nM IGF-I) and 20 nM (right bars, Botox + 20 nM IGF-I).Note that under Botox, LTD is induced by 10 nM IGF-I and that LTP can be restored by increasing the IGF-I concentration to 20 nM.*p < 0.05, **p < 0.01, and ***p < 0.001; student's paired t-test.# p < 0.01, ## p < 0.01, and ### p < 0.001; one-way ANOVA with post hoc Holm-Sidak; n.s., nonsignificant (p > 0.05).

FIGURE 4
FIGURE 4 Spiking activity is able to induce LTD IGF1 without exogenous application of IGF-I.(A top) Superimposed representative EPSCs recorded before (1) and 40 min after (3) spiking activity evoked by synaptic stimulation increase (2, SSI) in normal ACSF (black trace, SSI) and under NVP (red trace, SSI + NVP).(A bottom) Time course of the EPSC recorded before, during and after SSI in control (SSI, black circles), under NVP (SSI + NVP, white circles) and under Botox (SSI + botox, red circles).(B left) Representative spiking activity evoked by the synaptic stimulation increase (2).(B right) Bar plot showing the EPSC peak amplitude as percentage of the control after SSI in ASCF (SSI, black bar), under NVP (SSI + NVP, white bar) and under Botox (SSI + Botox, red bar).Note that SSI is able to induce an LTD that is prevented by both NVP and Botox and then is very similar to LTD IGF1 .(C top) Superimposed representative EPSCs recorded before (1) and 40 min after (3) spiking activity evoked by simulated spike (2, simulated spike) in normal ACSF (black trace, stimu.spike) and under NVP (red trace, stimu.spike + NVP).(C bottom) Time course of the EPSC recorded before, during and after SSI in control (SSI, black circles), under NVP (SSI + NVP, white circles).(D left) Representative simulated spike (2).(D right) Bar plot showing the EPSC peak amplitude as percentage of the control after simulates spike in ASCF (stimulat.Spike, purple bar), under NVP (stimulat.spike + NVP, blue bar).Note that simulated spikes are able to induce an LTD that is prevented by NVP similar to LTD IGF1.(E) Plot of the variance (1/CV 2 , where CV is coefficient of variation) as a function of the mean peak EPSC amplitude during simulated spike protocol normalized to control conditions (Mr).Note that the values were grouped following the diagonal suggesting that LTD was due to a change in presynaptic release properties.*p < 0.05, **p < 0.01, and ***p < 0.001; student's paired t-test.# p < 0.01, ## p < 0.01, and ### p < 0.001; one-way ANOVA with post hoc Holm-Sidak; n.s., nonsignificant (p > 0.05).
) synapses.Moreover, Double immunocytochemical analysis for pre (vGAT)-and post-(PSD95) synaptic markers and IGF-IR indicate the presence of immunopositive puncta opposing each other in close proximity (Figures 5C,D), in agreement with the presence of IGF-IR in both pre-and post-synaptic terminals.The presence of IGF-IR in the presynaptic terminals of both glutamatergic and GABAergic synapses supports the presynaptic LTD of the EPSCs induced by 7 nM IGF-I (present results) and the presynaptic LTD of the IPSCs generated by 10 nM IGF-I (Noriega-Prieto et al., 2021).

FIGURE 5 IGF-
FIGURE 5 IGF-IRs are present at presynaptic sites of both excitatory and inhibitory synapses.Electron micrographs demonstrating IGF-IR silver enhanced immunogold labeling (white arrows) within both excitatory (A) and inhibitory (B) synapses.Immuno-particles were found associated to presynaptic terminals (A,B).Asterisks indicate presynaptic terminals.Scale bars, 0.2 μm.Apposition of IGF-IR immunoreactive puncta (red) with both the postsynaptic marker PSD95 (green, white arrowheads in C) and the presynaptic marker vGAT (green, white arrowheads in D) agree with localization of IGF-IR at both sides of the synapse.DAPI staining of cell nuclei (blue).

FIGURE 6
FIGURE 6 Model showing the concentration dependent actions of IGF-I.Simplified cartoon summarizing the novel results revealed in this work.(Left part) Under high extracellular levels of IGF-I (between 10 and 20 nM), occurring during high activity at pyramidal neurons (PN) of the barrel cortex during processing information of active whisker, LTP IGF1 is induced what favors Hebbian LTP and the associative learning and memory.Note that high extracellular IGF-I levels during spiking activity at barrel cortex are achieved by vesicular release of IGF-I stored inside PNs and maybe because facilitated IGF-I entry from plasmatic IGF-I.Indeed there is an increase in the uptake of IGF-I by the brain during whisker stimulation that correlates with frequency-dependent changes in cerebral blood flow in the barrel cortex (Nishijima et al., 2010).(Right part) On the contrary during low activity of barrel cortex PNs during resting whisker, both vesicular release of IGF-I and IGF-I entry from plasmatic levels should be lowered and then a lower IGF-I extracellular concentration is expected (below 10 nM) and LTD IGF1 is induced what impairs Hebbian LTP and the associative learning and memory (scheme was performed with BioRender).